Toyota has partnered with experts from the University of Michigan’s Mcity to rethink the parameters of autonomous driving and establish the concept of “roadmanship” between autonomous and human-driven cars.
Toyota has recently expanded investments to its Collaborative Safety Research Center (CSRC) to take a deeper look into the safety needs of future mobility by researching three aspects: safety assurance, human-centric, and assessment. This latest collaboration with Mcity aims to examine everyday driving situations to specify what “normal” is for the regular driver.
“We want vehicles that play well with others, not creating new potential safety hazards,” said John Lenneman, a senior principal research scientist at CSRC. “How to measure that and compare it to human driving remains an open question.”
Autonomous cars are safer, but are they creating repugnance among human drivers? Tech developers began reassessing the parameters of autonomous driving a few years ago when the first vehicles started testing on public roads. Early research points to autonomous vehicles as potential hazards to other vehicles on the road because they clogged up traffic, said Toyota.
This concept has led to the need to examine roadmanship, or the idea of displaying a standard range of behaviors when autonomous cars are on the roads or highways. “As we start implementing vehicle automation, we’re evolving our thinking around safety to consider human behavior,” Lenneman added. “It’s also how well they (AVs) interact with road users.”
The first study examined how much of a gap human-driven vehicles require when turning left against oncoming traffic. The CSRC and Mcity surveyed an intersection in Jackson Hole, Wyoming, using video cameras running for 24 hours. After gathering hours of footage and more than 5,000 driving situations (in all weather conditions, including rain and snow), the team converted the data into a birds-eye-view orientation and utilized machine learning to measure acceptable and unacceptable gaps.
Findings suggest that human drivers use distance to decide whether it’s safe or unsafe to complete the turn. Researchers concluded that drivers judge what they see instead of calculating the speed of oncoming traffic or determining the time it takes to make the turn.
The second study looked into how drivers interact on a five-point roundabout. According to the CSRC, studies show that drivers entering the circles of moving traffic focus on the angle of approaching cars and trucks rather than determining the distance. Furthermore, the findings show that drivers are making a geometric assessment of the trajectory of oncoming vehicles to determine if it’s safe to proceed. “The challenges don’t end with other vehicles,” said Greg Mcguire, managing director of Mcity. “Human drivers must also respond to and exert more caution around other people riding bikes, motorcycles, and walking. Autonomous vehicles will need to do the same.”
The CSRC adds the long-term goal of AV roadmanship studies is to create a more harmonious relationship between autonomous vehicles and human drivers. In addition, the data helps development engineers to quantify driving specifications when testing self-driving cars.